On April 3rd, I asked for some help with a survey and an astonishing sixty six of you took part. In fact this was not a survey so much as an experiment aimed at teasing out a bit of evidence relating to that oft-quoted rule regarding the number of characters a short story can accommodate [three, evidently]. In a recent tutorial piece of around 5000 words, I had four characters and I was challenged to consider whether I could lose one by giving his actions to another. Well, I rather liked ‘Eric’ even though [maybe because] he is a vulnerable little weasel who takes out on others the injustices he has experienced himself. I thought he carried an important theme in the story that the other character could not. I also thought that, as the two characters pretty much operated as a pair throughout, they might actually function as one unit as far as memorability was concerned. I put this point, along with the proviso that I had no evidence for it, and I was challenged to find some.

And this is where we step away a little from literature, which might not be quite the approach my tutor was expecting! To me, evidence is not someone else’s opinion, whatever their status and however well argued. Evidence is material generated as the result of  a question posed within a particular methodological framework. It is much the same theoretical shift as the appreciation that there are two kinds of research: the sort that involves looking up facts and details set out by others, and the sort that tests hypotheses in order to find a new platform by which to understand something. The first kind usually informs the second and also exists as a valid entity in its own right, but the second addresses beliefs and sets out to find evidence for them.

Or actually it doesn’t. The really important thing about research is that it sets out to find disconfirmatory evidence (cf Karl Popper), not just evidence that supports the theory. In other words, if I want to demonstrate that unicorns don’t exist (I know, that’s contentious), I have to be sure my study will give unicorns every possible opportunity to appear and not just look in my enclosed back yard with its unicorn repellent wood stained fencing. For this study, my unicorn was the idea that characters do not operate as a single unit and so I needed to design the survey so that this would show up if it were true. This meant putting up a challenge that looked like one thing but was targeting something else so that participants could not know what I thought might happen.  I hoped most people would take this is a straight remembering task and not drill down to the real purpose.

The underlying theory for my belief about paired characters comes from cognitive psychology in which studies of memory show that, if we can ‘chunk’ information into handy groups, we will recall more items from a list than if we can’t. Sometimes we do this semantically – remembering all the vegetables or all the colours, sometimes rhythmically as in telephone numbers. Repetition has a role in that the more we use a particular string of data, the more it becomes a single rather than a multiple unit. Single units take up less space in memory – they constitute less of a cognitive load – so I wondered, if two characters operate consistently as a pair, are they represented in memory as a single unit, thereby allowing them to count as one for short story purposes and not confuse the heck out of people?

When I think of Eric Morecambe I don’t have to think separately about Ernie Wise – I just think ‘Morecambe and Wise’ almost as one word. Similarly ‘Mork and Mindy’ even though I have no idea why I know those names. My suspicion is that they are, cognitively, one and not two units of information. Other characters, Poirot and Sherlock for instance, are not pairs even though they are contextually similar and so I doubted they would be stored as a single unit. If then, people saw a list comprising a mix of half of these pairs and non-pairs and then had to decide from a second list which ones they had just seen, would they check any of the other halves of those pairs and non-pairs and if so, would there be any difference between those two groups? I thought that a) they would ‘recognise’ names on the second list that had not been on the first list, and b) that more of these would be from the paired group than the non-paired group.

The next part of the study was a little more difficult. I asked people to read another list of names – these were halves of pairs and non-pairs that they had not seen already – and then see how many they could recall. Recall is a harder task than recognition because there are no cues to trigger a memory and so people would need to search their minds to respond. I thought again that there would be names in the response lists that had not been on the list people read, and that more of these would be from the paired than the non-paired group.

Here’s what happened: How many characters can a short story accommodate 2 pdf

I want to say a very big thank you to everyone who took part and to assure you all that you are pretty darned normal – at least as far as recognising and recalling stuff is concerned.  For the rest of your personality and behaviour – you’re on your own with that!